Structural Evolution of Gold Nanorods during Controlled Secondary

Aug 22, 2007 - Structural Evolution of Gold Nanorods during Controlled Secondary Growth. Heidrun A. Keul,Martin Möller,* andMichael R. Bockstaller*. ...
0 downloads 11 Views 803KB Size
Langmuir 2007, 23, 10307-10315

10307

Structural Evolution of Gold Nanorods during Controlled Secondary Growth Heidrun A. Keul,† Martin Mo¨ller,*,† and Michael R. Bockstaller*,‡ DWI and Institute of Technical and Macromolecular Chemistry, RWTH Aachen UniVersity, Pauwelsstr. 8, 52056 Aachen, Germany, and Department of Materials Science and Engineering, Carnegie Mellon UniVersity, 5000 Forbes AVenue, Pittsburgh, PennsylVania 15206 ReceiVed May 24, 2007. In Final Form: July 2, 2007 Single-crystalline gold nanorods synthesized by the Ag(I)-mediated seeded-growth method (see: El-Sayed, M. A.; Nikoobakht, B. Chem. Mater. 2003, 15, 1957) were used as seeds for the preferential overgrowth of gold on particular crystallographic facets by systematic variation of the conditions during overgrowth. The results support previous reports about the relevance of the cationic surfactant cetyltrimethylammonium bromide (CTAB) and Ag(I) in stabilizing anisotropic particle shapes and demonstrate that the regulation of the amount of ascorbic acid facilitates the preferential overgrowth of {111} crystal facets to form Ξ-type particle shapes. Interestingly, secondary overgrowth is found to inevitably result in a loss of particle shape anisotropy. A mechanism based on surface reconstruction is proposed to rationalize the “shape-reversal” that is generally observed in the nanorod growth process, that is, the initial increase and subsequent decrease of particle anisotropy with increasing reaction time. High-resolution electron microscopy analysis of gold nanorods reveals clear evidence for (1 × 2) missing row surface reconstruction of high energetic {110} facets that form during the initial phase during particle growth.

Introduction The intimate dependence of the physical properties on the length scales and geometry of metal nanoparticles has motivated research in developing methodologies for size- and shapecontrolled crystal growth. One example that has attracted particular interest is the synthesis of gold nanorods, in which the anisotropic particle shape gives rise to the splitting of the plasmon resonance in transverse and longitudinal modes. The sensitivity of the longitudinal plasmon mode on the particle’s anisotropy facilitates opportunities for a wide range of technological applications of gold nanorods such as sensing, imaging, or photothermal therapy technologies that have been investigated recently.1-7 The synthesis of anisotropic gold particles generally requires a mechanism for symmetry breaking to facilitate the anisotropic growth of face-centered cubic gold, and several wetchemical methodologies have been developed that are based on template-assisted growth or on the addition of structure-guiding additives. The first realization of gold nanowires was demonstrated by electrodeposition using porous membranes based on anodized alumina or etched polycarbonate as structure-guiding templates.8-14 While this technique has the advantage of being compatible with a wide variety of material compositions, the * To whom correspondence should be addressed. E-mail: Bockstaller@ cmu.edu (M.R.B.); [email protected] (M.M.). † RWTH Aachen University. ‡ Carnegie Mellon University. (1) El-Sayed, M. A. Annu. ReV. Phys. Chem. 2003, 54, 331. (2) Hutter, E.; Fendler, J. H. AdV. Mater. 2004, 16, 1685. (3) Burda, C.; Chem, X.; Narayanan, R.; Xia, Y.; Halas, N. J. MRS Bull. 2005, 30, 338. (4) Haes, A. J.; Haynes, C. L.; McFarland, A. D.; Schatz, G. C.; Van Duyne, R. P.; Zou, S. MRS Bull. 2005, 30, 368. (5) Liz-Marza´n, L. M. Langmuir 2006, 22, 32. (6) Murphy, C. J.; Sau, T. K. Langmuir 2004, 20, 6414. (7) Huang, X.; El-Sayed, I. H.; Qian, W.; El-Sayed, M. A. J. Am. Chem. Soc. 2006, 128, 2115. (8) Possin, G. E. ReV. Sci. Instrum. 1970, 41, 772. (9) Martin, C. R. Chem. Mater. 1996, 8, 1739. (10) Scho¨nenberger, C.; van der Zande, B. M. I.; Fokkink, L. G. J.; Henry, M.; Schmid, C.; Kru¨ger, M.; Bachtold, A.; Huber, R.; Birk, H.; Staufer, U. J. Phys. Chem. B 1997, 101, 5497. (11) Cepak, V. M.; Martin, C. R. J. Phys. Chem. B 1998, 102, 9985.

dimensions of the attained nanostructures are typically 1 order of magnitude larger than those of alternate wet-chemical techniques and only applicable to the technology areas mentioned above. Moreover, membrane imperfections have resulted in significant numbers of more complex particle shapes that cannot be readily separated from the product.15 Esumi et al. were the first to use photochemical reduction in the presence of cylindermicelle forming cationic surfactants (cetyltrimethylammonium chloride) to attain anisotropic gold nanostructures.16 This approach demonstrated the feasibility of anisotropic growth in precipitation reactions; the resulting threadlike gold nanostructures, however, exhibited poor morphological control. More recently, the Murphy group introduced a seeded-growth approach in which spherical gold nanocrystals are used as seeds to grow gold nanorods in the presence of weak reducing agents and cationic surfactants.17,18 This method has yielded gold nanorods with aspect ratios ranging from 2 to 10 and average rod diameters of some tens of nanometers and has attracted enormous interest as a scalable synthetic technique for well-defined gold nanorods. El-Sayed and coworkers demonstrated that (similar to the electrochemical growth technique) the efficiency of the reaction as well as the control of nanorod anisotropy can be enhanced by the presence of Ag(I) during rod formation.19-22 Despite significant efforts in under(12) Van der Zande, B. M. I.; Bohmer, M. R.; Fokkink, L. G. J.; Scho¨nenberger, C. Langmuir 2000, 16, 451. (13) Nicewarner-Pena, S. R.; Freeman, R. G.; Reiss, B. D.; He, L.; Pena, D. J.; Walton, I. D.; Cromer, R.; Keating, C. D.; Natan, M. J. Science 2001, 294, 137. (14) Thurn-Albrecht, T.; Schotter, J.; Ka¨stle, G. A.; Emley, N.; Shibauchi, T.; Krusin-Elbaum, L.; Guarini, D.; Black, C. T.; Tuominen, M. T.; Russell, T. P. Science 2000, 290, 2126. (15) Scho¨nenberger, C.; van der Zande, B. M. I.; Fokkink, L. G. J.; Henry, M.; Schmid, C.; Kru¨ger, M.; Bachtold, A.; Huber, R.; Birk, H.; Staufer, U. J. Phys. Chem. B 1997, 101, 5497. (16) Esumi, K.; Matsuhisa, K.; Torigoe, K. Langmuir 1995, 11, 3285. (17) Jana, N. R.; Gearheart, L.; Murphy, C. J. AdV. Mater. 2001, 13, 1389. (18) Jana, N. R.; Gearheart, L.; Murphy, C. J. J. Phys. Chem. B 2001, 105, 4065. (19) El-Sayed, M. A.; Nikoobakht, B. Chem. Mater. 2003, 15, 1957. (20) Gole, A.; Murphy, C. J. Chem. Mater. 2004, 16, 3633. (21) Perez-Juste, J.; Liz-Marzan, L. M.; Carnie, S.; Chan, D. J. C.; Mulvaney, P. AdV. Funct. Mater. 2004, 14, 571.

10.1021/la7015325 CCC: $37.00 © 2007 American Chemical Society Published on Web 08/22/2007

10308 Langmuir, Vol. 23, No. 20, 2007

standing the mechanism of rod growth, the role of auxiliary reagents present during the reaction and even the structure of the nanorods are still disputed. This can be attributed to the complex reactant composition and the sensitivity of crystal growth to the reaction conditions. In the following, the major conclusions of previous studies of the structure of gold nanorods will be summarized to provide the context for our work. Using combined high-resolution electron microscopy (HRTEM) and selected area diffraction (SAD), Johnson et al. suggested that the nanorods exhibit a 5-fold multitwinned structure with 5 {100} and 10 {111} facets similar to those of nanowire structures that were previously grown from gold, copper, and silver.23-25 In this model, the nanorod is assumed to evolve from a 5-fold twinned decahedral seed-crystal by preferential growth along the [110] direction, with the capping agent directing the structure evolution by stabilizing the generated {100} facets rather than the low-energy {111} facets. Interestingly, the authors found the seed-crystals to be single-crystalline, suggesting that the twinning process occurs through particle growth or coagulation at the early stages of the growth process. More recently, Liu et al. studied the implications of seed-crystal morphology on nanorod formation. Whereas for twinned seed-crystal structures the formation of 5-fold twinned bipyramids was observed, the authors report the formation of single-crystalline gold nanorods if singlecrystalline seed-crystals are being used to nucleate rod formation.26 In this case, side facets were found to be of the type {100} or {110} with {111} facets forming the side facets along the head and tail faces of the rod. These findings are in agreement with an independent report by Song et al. who found gold nanorods to be single-crystalline when synthesized following the procedure used in the Ag(I) assisted seeded-growth technique.27 A comment to be made is that the multitwinned nanorods studied by Johnson et al. were about twice in thickness (d ) 30-40 nm) than the nanorods studied by Liu et al. and Song et al. (d ) 15 nm) and were prepared in the absence of Ag(I). From these previous studies, it appears that there is not a unique mechanism responsible for the anisotropic growth, and the role of the respective reaction parameters remains elusive. In this contribution, we present a sequential growth procedure of gold nanorods that provides insight into the morphological effects of the various reactants present in the growth solution. In a first step, gold nanorods are synthesized using the Ag(I) mediated seeded-growth procedure as described in ref 19. The rods are then purified from spherical particle byproducts and excess surfactant by centrifugation and washing and subsequently applied as “seed-rods” in a secondary growth solution that has been modified by systematic deletion or concentration of reactants from the original growth solution. The procedure is depicted in Scheme 1. Conceptually, the experimental approach presented in our paper is most closely related to the overgrowth procedure reported by Song et al.; however, the focus of the present work is to provide a systematic analysis of the implications of the overgrowth conditions on the morphology of gold nanorods using well-defined seed-nanorods as a reference state as well as an analysis of the time dependence of rod formation.27 The distinct morphological features that result from the overgrowth under (22) Murphy, C. J.; Sau, T. K.; Gole, A.; Orendorff, C. J. MRS Bull. 2005, 30, 349. (23) Johnson, C. J.; Dujardin, E.; Davis, S. A.; Murphy, C. J.; Mann, S. J. Mater. Chem. 2002, 12, 1765. (24) Lisiecki, L.; Filankembo, H.; Sack-Kohngel, K.; Weiss, K.; Pileni, M. P.; Urban, J. Phys. ReV. B 2000, 61, 4968. (25) Hofmeister, H.; Nepijko, S. A.; Ievlev, D. N.; Schulze, W.; Ertl, G. J. Cryst. Growth 2002, 234, 773. (26) Liu, M.; Guyot-Sionnest, P. J. Phys. Chem. B 2005, 109, 22192. (27) Song, J. H.; Kim, F.; Kim, D.; Yang, P. Chem.sEur. J. 2005, 11, 910. (28) Iijima, S.; Ichihashi, T. Phys. ReV. Lett. 1986, 56, 616.

Keul et al. Scheme 1. Illustration of the Sequential Synthesis Procedure and the Resulting Particle Geometries of Gold Nanorods after Overgrowth

modified reaction conditions give insight into the particular relevance of the surfactant, Ag(I), and the reducing agent in stabilizing anisotropic particle growth. Next to providing valuable information about the mechanism of rod formation, the sequential growth approach is shown to facilitate excellent control over particle size and shape, which is particularly interesting for the synthesis of technologically relevant low-anisotropy nanocrystals that are difficult to attain in single step procedures. Materials and Methods Materials. Tetrachloroauric(III) acid trihydrate (∼52% as Au) and cetyltrimethylammonium bromide (CTAB) were purchased from Fluka. Sodium borohydride, L-(+)-ascorbic acid, and silver nitrate were purchased from Aldrich. All chemicals were used as received. Bidistilled water was used in all syntheses (Elga Purelab Ultra Plus UV). Nanorod Synthesis. Nanorods were prepared via a seed mediated growth procedure on the basis of the reports published by El-Sayed et al.19 In typical nanorod synthesis, a growth solution was prepared by mixing aqueous silver nitrate stock solution (125 µL, 1 µmol), CTAB stock solution (4 mL, 0.8 mmol), bidistilled water (4.625 mL), and a stock solution of hydrogen tetrachloroauric(III) acid trihydrate (1.4 mL, 4.2 µmol) in the given order and then stirring at room temperature. The yellow solution turned colorless upon addition of aqueous ascorbic acid solution (90 µL, 4.5 µmol). A seed solution was prepared by mixing CTAB stock solution (5 mL, 1 mmol), bidistilled water (4.2 mL), and a stock solution of hydrogen tetrachloroauric(III) acid trihydrate (0.8 mL, 2.4 µmol) in the given order. Under vigorous stirring, freshly prepared, ice-cold, aqueous sodium borohydride solution (0.6 µL, 6 µmol) was added and the solution turned from orange to yellow-brown. The seed solution (12 µL) was added to the growth solution to start nanorod growth. The seed solution was used within 5 min up to 2 days after preparation. After about 2.5 h of reaction time, the nanorod solution was centrifuged at 11 000 rpm for 30 min. The supernatant was removed.

Structural EVolution of Gold Nanorods After dilution of the concentrate, centrifugation was repeated to remove excess surfactant from the solution. Unmodified Nanrods. The rod concentrate, obtained via centrifugation, was diluted with 10 mL of bidistilled water. Nanorod Modification with Regular Growth Solution. A growth solution was prepared in analogy to the one used for nanorod synthesis. The as-prepared rod concentrate, obtained via centrifugation, was added to this second growth solution. Nanorod Modification with Growth Solution without Silver Nitrate. This modified growth solution was prepared by mixing CTAB stock solution (4 mL, 0.8 mmol), bidistilled water (4.750 mL), a stock solution of tetrachloroauric(III) acid trihydrate (1.4 mL, 4.2 µmol), and aqueous ascorbic acid solution (90 µL, 4.5 µmol) in the given order and then stirring at room temperature. To this second growth solution, the as-prepared rod concentrate, obtained via centrifugation, was added. Nanorod Modification with Growth Solution without CTAB. Bidistilled water (8.625 µL), aqueous silver nitrate stock solution (125 µL, 1 µmol), and a stock solution of tetrachloroauric(III) acid trihydrate (1.4 mL, 4.2 µmol) were mixed and added to the asprepared rod concentrate obtained via centrifugation. Subsequently, ascorbic acid solution (90 µL, 4.5 µmol) was added. Nanorod Modification with Growth Solution with an Excess of Ascorbic Acid. This modified growth solution was prepared by mixing CTAB stock solution (4 mL, 0.8 mmol), bidistilled water (4.750 mL), a stock solution of tetrachloroauric(III) acid trihydrate (1.4 mL, 4.2 µmol), and aqueous ascorbic acid solution (150 µL, 7.5 µmol) in the given order and then stirring at room temperature. To this second growth solution, the as-prepared rod concentrate, obtained via centrifugation, was added. Instrumentation. Absorption spectra were acquired using a Shimadzu UV/vis spectrophotometer (UV160A). TEM images were acquired using a Philips CM10 PW6020/10 transmission electron microscope with an accelerating voltage of 100 kV. TEM grids were prepared by putting a drop of the concentrated nanorod sample (30 µL) on a holey carbon or Formvar-coated and carbon-sputtered copper TEM grid, placing the grid on a filter paper, and evaporating the solution at room temperature. For each sample, a minimum of 100 particles were measured to obtain the average dimensions and the size distribution. For high-resolution transmission electron microscopy (HRTEM) and energy-dispersive X-ray (EDX) measurements, a FEI Tecnai F20 transmission electron microscope was used, containing an energy filter (Gatan GIF 2000), an EDX system (EDAX), and a HAADF detector.

Results and Discussion The synthesis of gold nanorods via the seeded-growth approach is carried out by combining an aqueous seed solution containing presynthesized spherical gold particles stabilized by CTAB with an aqueous solution of Au(I) that is stabilized by an excess amount of CTAB. The seeding process is performed by reduction of Au(III) with a strong reducing agent, resulting in fast nucleation and the formation of spherical nanocrystals. Subsequently, rod formation is facilitated by the reduction of Au(I) with a weak reducing agent, probably mediated by the catalytic effects of the metal particle surface. In the following, results of the structural characterization of the products of the respective stages will be presented. Seed Nanocrystals. Seed nanocrystals were found to be of an average particle diameter dSEED ) 4 ( 1 nm. Figure 1 depicts a HRTEM image of the seed-crystals, confirming their singlecrystalline structure. Further quantitative analysis of the seedcrystals reveals that about 75% are single-crystalline and about 25% exhibit a twinned structure that could in part be interpreted as a 5-fold twinned decahedral shape. This result agrees with reports by Song et al. and Liu et al. who also observed mostly single-crystalline seed-crystal morphologies when crystals were prepared following the procedure developed by El-Sayed et al.19,26,27

Langmuir, Vol. 23, No. 20, 2007 10309

Figure 1. High-resolution electron micrograph revealing the singlecrystalline structures of seed nanocrystals. Scheme 2. Illustration of Truncated Octahedra Representing Equilibrium Particle Shapes (Wulff Shape) of Golda

a See text for details. a-c correspond to an increasing importance of the {100} facets.

For crystalline materials, the equilibrium structure is generally predicted by the Wulff construction that aims to minimize the surface energy of a free particle. For gold, the surface energy for the low-crystallographic planes is γ{110} > γ{100} > γ{111}, and the Wulff polyhedron is equal to a truncated octahedron, exhibiting eight {111} and six {100} faces as indicated in Scheme 2.29,30 However, based on numerous studies of small gold nanocrystals, it is understood that the different crystal structures are separated by only small energy barriers (∼kBT), and thus, precipitation reactions at room temperature often result in the distribution of particle shapes.31,32 In particular, for small particle sizes, it was shown that the 5-fold twinned decahedral structure can be energetically favored, suggesting that the mixture of crystal morphologies is rather natural for the given particle size range.33 Because of the mostly single-crystalline morphology of the resulting gold nanorods (see next section), we conclude that single-crystalline gold nanocrystals indeed act as seeds for subsequent rod formation, although it cannot be excluded that single-crystalline nanocrystals undergo a twinning transformation (or vice versa) at the early stages after addition of growth solution. Primary Nanorod Formation. Nanorods were synthesized by the Ag(I)-mediated seeded-growth procedure and purified by centrifugation and separation of spherical particle byproducts as well as excess CTAB.19 Figure 2a depicts an electron micrograph of the purified rod solution that was subsequently subdivided into different batches for the overgrowth reactions. The particle size distribution shown in Figure 2b was evaluated from image analysis: the average anisotropy and rod thickness were determined to be 〈L〉/〈d〉 ) 3.4 ( 0.6 and 〈d〉 ) 11 ( 2 nm (with 〈L〉 and 〈d〉 denoting the average rod length and thickness, respectively) and confirmed by comparison of the experimental and theoretical longitudinal plasmon resonance frequencies (not shown here). After extraction and purification, the nanorods were (29) Ino, S. J. Phys. Soc. Jpn. 1969, 27, 941. (30) Marks, L. D. Rep. Prog. Phys. 1994, 57, 603. (31) Smith, D. J.; Petford-Long, A. K.; Wallenberg. L. R.; Bovin, J.-O. Science 1986, 233, 872. (32) Liu, H. B.; Ascensio, J. A.; Perez-Alvarez, M.; Yacaman, M. J. Surf. Sci. 2001, 491, 88. (33) Yacaman, M. J.; Ascencio, J. A.; Liu, H. B.; Gardea-Torresdey, J. J. Vac. Sci. Technol., B 2001, 19, 1091.

10310 Langmuir, Vol. 23, No. 20, 2007

Figure 2. (a) Bright-field electron micrograph depicting purified seed-nanorods. (b) Distribution of particle anisotropy determined by image analysis. The average particle anisotropy is 〈L〉/〈d〉 ) 3.4 ( 0.6, and the average rod thickness is 〈d〉 ) 11 ( 2 nm. (c) UV/vis absorption spectrum of a purified nanorod solution after 7 days (black solid line) and 28 days (red dashed line).

found to be stable in distilled water for weeks as evidenced by the almost unchanged absorption characteristics depicted in Figure 2c. A HRTEM image of a typical rod crystal structure and the different crystal orientations are illustrated in Figure 3. The nanorods were found to be mostly single-crystalline with {110} and/or {100} side facets and {100} and {111} facets in the endcap regions of the rods. This is in agreement with a series of previous studies that reported single-crystalline nanorod morphologies.26,27,34,35 The preferential growth of the {110} and {100} planes has been attributed to kinetic effects resulting from the preferential adsorption of the surfactant on the low-density crystal facets due to their increased surface energy, and because of the more favorable epitaxial relation.36 We note that the occurrence (34) Wang, Z. L.; Mohamed, M. A.; Link, S.; El-Sayed, M. A. Surf. Sci. 1999, 440, L809. (35) Kou, K. S.; Zhang, S. Z.; Tsung, C.-K.; Yeung, M. H.; Shi, Q. H.; Stucky, G. D.; Sun, L. D.; Wang, J. F.; Yan, C. H. J. Phys. Chem. B 2006, 110, 16377. (36) Murphy, C. J.; Sau, T. K.; Gole, A. M.; Orendorff, C. J.; Gao, J.; Guo, L.; Hunyadi, S. E.; Li, T. J. Phys. Chem. B 2005, 109, 13857.

Keul et al.

Figure 3. High-resolution electron micrographs of a seed nanorod confirming the growth of nanorods along the [100] direction and proposed rod morphology. (a) HRTEM image of a nanorod revealing {100} planes normal to the growth direction. (b) HRTEM image of a nanorod revealing {111} crystal orientation. Side facets are constituted by low-density {100} and {110} crystal planes, and end facets are constituted of {111} and {100} facets. (c) Illustration of the proposed rod morphology (rod-end morphology speculative).

of high-energy {110} planes is particular in the formation of single-crystalline gold nanorods and might be key in understanding some of the kinetic phenomena observed in the synthesis of gold nanorods (see below). It is instructive to consider the distribution of bromide counterions in the final nanorod solution. Figure 4 depicts the gold and bromine distribution determined by energy-dispersive X-ray analysis along the perimeter of three parallel rods that were cast from the purified solution on holey carbon-coated copper grids. While the “rippling” of the bromine concentration needs to be interpreted with caution since the measured signal is affected by the finite resolution of the probe beam (∼8 nm), the rapid vanishing outside the nanorod (at distance X > 23.5 nm) indicates that bromine is accumulated in the immediate vicinity of the gold surface. This is in support of previous arguments that bromine participates in the stabilization of crystal facets or mediates the binding of surfactants to the crystal surface.6,37 To better understand the relevance of the respective reaction parameters, equal batches of the purified nanorod solution were (37) Chang, S. S.; Shih, C. W.; Chen, C. D.; Lai, W. C.; Wang, C. R. Langmuir 1999, 15, 701.

Structural EVolution of Gold Nanorods

Langmuir, Vol. 23, No. 20, 2007 10311

Figure 5. (a) Bright-field electron micrograph of gold nanorods after 3 h of overgrowth using regular growth solution (i.e., growth solution equal to conditions used in ref 19) revealing particles with reduced anisotropy and retained morphology. The inset shows a magnified area. (b) Distribution of particle anisotropy determined by image analysis. The average particle anisotropy is 〈L〉/〈d〉 ) 2.4 ( 0.3, and the average rod thickness is 〈d〉 ) 17 ( 2 nm. Figure 4. (a) Scanning transmission electron micrograph of purified seed nanorods deposited on holey carbon-coated Cu grids. The red line indicates the trace for elemental analysis by energy-dispersive X-ray analysis (EDX). (b) EDX spectrum of purified seed nanorods depicting the spatial concentration of gold (yellow curve) and bromine (green curve). Bromine is concentrated in the immediate vicinity of the particle surface.

exposed to a subsequent overgrowth process with modified growth solutions and with the systematic variation of one reaction component. Morphology of Nanorods After Sequential Overgrowth. 1. Regular Growth Solution. After exposing the primary nanorods to a second regular growth solution (i.e., the composition of the growth solution is unchanged), rather isotropic overgrowth of the nanorods is observed, resulting in an increased particle volume (〈d〉 ) 17 ( 2 nm) but a decreased anisotropy of 〈L〉/〈d〉 ) 2.4 ( 0.3. Figure 5 depicts electron micrographs of the resulting nanorods after 3 h of reaction time along with the corresponding particle size distribution. The number of spherical particles was found to be unchanged, indicating that the nanorods effectively acted as seeds for the overgrowth. These results are in good agreement with earlier studies that reported a decrease in particle anisotropy after further addition of growth solution during rod synthesis.19 Decreased particle anisotropy suggests the growth of {111} planes even though the reaction conditions are about equal to the ones that originally favored the growth of highenergy facets. A possible origin for this “shape-reversed” growth

kinetics will be discussed below in the context of time effects on rod formation. 2. Absence of Ag(I). Figure 6 depicts electron micrographs after exposing the primary rod particles to a modified growth solution devoid of Ag(I) after 3 h of reaction time. Throughout the sample, rather uneven particle shapes were observed that suggest isotropic overgrowth of the particles. The average particle anisotropy of 〈L〉/〈d〉 ) 2.3 ( 0.4 and average width of 〈d〉 ) 24 ( 6 nm were determined from the size distribution shown in Figure 6b. The dumbbell-type shape of the particles after overgrowth indicates a preference for growing {111} facets. The relevance of Ag(I) to induce anisotropic particle growth is particularly striking at longer reaction times after which no anisotropic particle shapes could be detected (see next section). The relevance of Ag(I) in regulating the growth rates along the various crystallographic planes has been discussed by several authors. Liu et al. demonstrated the stabilization of high-energy {110} and {100} facets in the presence of Ag(I) in the seeded growth of gold nanorods and rationalized their observation as a consequence of the stabilization of high-energy gold crystal facets by underpotential deposition of a silver monolayer.26 A similar argument was made to rationalize the shape-regulating effect of Ag(I) on the growth of electrochemically deposited gold nanowires.37 It is important to note that the implications of Ag(I) additives on rod formation might be different for singlecrystalline and multitwinned nanorod morphologies. Whereas

10312 Langmuir, Vol. 23, No. 20, 2007

Figure 6. (a) Bright-field electron micrograph of gold nanorods after 3 h of overgrowth using modified growth conditions 1 (i.e., absence of Ag(I) in the growth solution) revealing early stages of isotropic overgrowth. The inset shows a magnified area. (b) Distribution of particle anisotropy determined by image analysis. The average particle anisotropy is 〈L〉/〈d〉 ) 2.3 ( 0.4, and the average rod thickness is 〈d〉 ) 24 ( 6 nm.

for single-crystalline nanorods a symmetry breaking event (such as preferential adsorption of surfactant on high-energy facets) is required to induce anisotropic growth, the presence of stabilizing agents might be less stringent for multitwinned seed-crystal morphologies. For example, it has been argued that the introduction of twin planes in 5-fold twinned decahedral nanocrystals (that are considered to be the seed-crystals that give rise to 5-fold twinned nanorod morphologies) favors anisotropic growth through recurring twinning events and defect formation.38,39 3. Absence of CTAB. The overgrowth on purified gold nanorods in the absence of surfactant (CTAB) was tested by addition of Au(III) and Ag(I) in concentrations equal to those of the original growth solution and subsequent addition of the reducing agent (ascorbic acid). Concentrations were chosen such as to prevent secondary nucleation and to avoid precipitation. Figure 7 depicts electron micrographs and the size distribution of the resulting gold nanorods 3 h after addition of ascorbic acid. Consistently, the particle shapes were found to be rodlike with a decreased anisotropy of 〈L〉/〈d〉 ) 2.0 ( 0.3 and an average rod thickness of 〈d〉 ) 32 ( 5 nm. The decrease in rod anisotropy was confirmed by the blue-shift of the longitudinal plasmon frequency (not shown here). Furthermore, analysis of the electron micrographs suggests that the number of spherical nanocrystals essentially (38) Lofton, C.; Sigmund, W. AdV. Funct. Mater. 2005, 15, 1197. (39) Elechiguerra, J. L.; Reyes-Gasga, J.; Yacaman, M. J. J. Mater. Chem. 2006, 16, 3906.

Keul et al.

Figure 7. (a) Bright-field electron micrograph of gold nanorods after 3 h of overgrowth using modified growth conditions 2 (i.e., absence of CTAB in the growth solution) revealing the reduced anisotropy and pronounced volume increase of overgrown nanorods, indicating a preferential overgrowth of {100} and {110} facets. The inset shows a magnified area. (b) Distribution of particle anisotropy determined by image analysis. The average particle anisotropy is 〈L〉/〈d〉 ) 2.0 ( 0.3, and the average rod thickness is 〈d〉 ) 32 ( 5 nm.

remained constant during the reaction, indicating that growth occurred primarily on the seed-nanorods. Several conclusions can be drawn from these observations. First, the rate of increase of the particle volume significantly exceeds alternate growth conditions tested in our study, indicating that in the absence of CTAB overgrowth occurs at an accelerated rate. This is in agreement with previous reports of a deceleration effect of CTAB on particle nucleation and growth and is likely related to interactions between the solute species and the surfactant micelles.21 Second, the decrease in particle anisotropy indicates the preferred formation of {111} facets to the expense of higher energy facets. The role of the surfactant as well as its counterion as a shape-regulating parameter has been the subject of intense debate. Since CTAB is used in significant excess quantities, its role will likely be complex, involving the coordination of anions, the stabilization of Au(I), the deceleration of the reaction rate and therefore the deposition rate, as well as the stabilization of the nanorods in solution. There is a fundamental agreement in the literature that one effect of the surfactant is the stabilization of nanorods by the formation of a surfactant bilayer “encapsulating” the rod.6,19,21,36 Based on a systematic study of the influence of alkyl trimethylammonium bromide surfactants on nanorod formation, it was concluded that the particular stabilization of surfactant layers through interdigitation gives rise to a minimum length of the alkyl residue that is necessary to afford rod formation.40 Furthermore, it was suggested that preferential (40) Rodriguez-Lopez, J. L.; Montejano-Carrizales, J. M.; Jose-Yacaman, M. Mod. Phys. Lett. 2006, 20, 725.

Structural EVolution of Gold Nanorods

Figure 8. (a) Bright-field electron micrograph of gold nanorods after 3 h of overgrowth using modified growth conditions 3 (i.e., excess of ascorbic acid in the growth solution) revealing Ξ-type particle shapes, indicating a preferential overgrowth of {111} facets. The inset shows a magnified area. (b) Distribution of particle anisotropy determined by image analysis. The average particle anisotropy is 〈L〉/〈d〉 ) 2.6 ( 0.3, and the average rod thickness is 〈d〉 ) 22 ( 3 nm.

binding of the ammonium headgroup to the high-energy crystal faces of the growing nanorods kinetically favors anisotropic crystal growth. More experimental and theoretical studies will be necessary to elucidate the role of the surfactant. 4. Excess Ascorbic Acid. The effect of excess amounts of reducing agent was tested by increasing the amount of ascorbic acid to 170% of the literature value.19 Figure 8 depicts an electron micrograph along with the corresponding particle size distribution after 3 h of reaction time. Particles were found to exhibit an average particle anisotropy of 〈L〉/〈d〉 ) 2.6 ( 0.3 and an average rod thickness of 〈d〉 ) 22 ( 3 nm, measured at the center of the rods. The overgrowth was found to be enhanced along the head/ tail regions of the nanorods, resulting in Ξ-type particle shapes with relative narrowing along the rod center region, indicating that overgrowth preferentially occurred along the exposed {111} facets. This confirms earlier results by Gou et al. who reported similar structures after addition of excess amounts of ascorbic acid to as-prepared nanorod solutions.42 The effect of ascorbic acid to promote the growth of {111} gold crystal facets was also reported by Elechiguerra et al. who demonstrated the formation of star-shaped nanocrystals by selective growth of the {111} faces of cuboctahedral seed-crystals after addition of ascorbic acid.39 The detailed mechanism for the promotion of the growth of high-density crystallographic planes by ascorbic acid is not understood; however, an important conclusion can be drawn (41) Gao, J.; Bender, C. M.; Murphy, C. J. Langmuir 2003, 19, 9065. (42) Gou, L.; Murphy, C. J. Chem. Mater. 2005, 17, 3668.

Langmuir, Vol. 23, No. 20, 2007 10313

Figure 9. UV/vis absorption spectra of sealed overgrowth solutions at intermediate and long reaction times: (a) 7 days and (b) 28 days. Curves represent regular growth conditions (blue dashed line) as well as growth solutions devoid of CTAB (red dotted line) and Ag(I) (black solid line). Unchanged positions of the transverse and longitudinal plasmon frequencies for growth conditions devoid of CTAB reveal constant particle anisotropy and confirm an increased rate of overgrowth. In the absence of Ag(I), a transition to spherical particles is observed at long reaction times (see text for details).

from the pronounced shape uniformity of the Ξ-type particles. Assuming a similar mechanism to be responsible for the growth of the various seed-nanorods, the uniformity of the particle morphologies after overgrowth confirms that the majority of rod morphologies should indeed be analogous to the rod morphologies depicted in Figure 3. Time Effects. In the discussion presented above, particle shapes were compared after a reaction time of 3 h which was chosen for experimental convenience. The influence of reaction time on particle evolution under secondary growth conditions was monitored by UV/vis absorption spectroscopy, which facilitates the determination of particle anisotropy for statistically representative particle ensembles via the relative positions of the transverse and longitudinal plasmon absorption frequencies. Figure 9 depicts the UV/vis absorption spectra of sealed solution samples of the secondary growth reactions involving regular growth solution and conditions devoid of CTAB or Ag(I) after a reaction time of 1 week and 4 weeks. The results confirm the short time observations; that is, a more pronounced blue-shift of the longitudinal plasmon absorption frequency (and thus a smaller particle aspect ratio) is observed for secondary growth conditions devoid of CTAB and Ag(I). For all samples, an increase in the absorption coefficient is observed after 4 weeks, indicating an increase in particle volume (the absorption coefficient of a particle is approximately proportional to its volume) as well as a blue-

10314 Langmuir, Vol. 23, No. 20, 2007

Keul et al. Scheme 3. Illustration of the (110) (1 × 2) Missing Row Surface Reconstruction Processa

Figure 10. UV/vis absorption spectra of a nanorod growth solution after t ) 60, 300, 600, 900, and 3600 s (lighter gray lines refer to longer reaction times). A shift of the plasmon absorption frequency indicates a change in the particle aspect ratio. Inset: rod aspect ratio calculated by a comparison of the theoretical and experimental absorption spectra (see ref 44 for details).

shift of the longitudinal plasmon frequency, indicating a decrease in particle anisotropy.43 For overgrowth conditions devoid of CTAB, the highest absorption coefficients are obtained in good agreement with the increased particle volumes determined by electron microscopy. After 4 weeks, particles obtained in the presence and absence of CTAB exhibit equal anisotropies as indicated by the congruent positions of the longitudinal plasmon absorbance frequencies (λL,+CTAB ) 619 nm, λL,-CTAB ) 613 nm). For secondary growth conditions devoid of Ag(I), the merging of transverse and longitudinal plasmon frequencies along with a pronounced increase in the amplitude of the transverse absorption band is observed, indicating the formation of large spherical particles. This observation further supports the relevance of Ag(I) in stabilizing low-density crystal facets that are associated with the formation of anisotropic particle shapes. 1. Interpretation of Time Effects on Rod EVolution. Interestingly, all secondary growth conditions were found to result in a decrease of the particle aspect ratios. The shape-reversal in the evolution of gold nanorods is a typical characteristic of rod formation: at short reaction times (the details depend on the reaction conditions), anisotropic growth is observed, whereas at longer reaction times the anisotropy is found to decrease, suggesting an “optimum reaction time” for the synthesis of high aspect ratio nanorods. The increase and subsequent decrease of the rods’ aspect ratio is evidenced in Figure 10 that depicts the time dependent UV/vis spectra of a regular growth solution during 1 h of reaction time. In these spectra, the particle aspect ratio can be deduced from the ratio of the peak wavelength of the longitudinal and transverse plasmon absorptions using a procedure that has been outlined in ref 44. Analysis of the optical properties of the nanorod solutions reveals that a maximum aspect ratio is obtained after about 7 min of reaction time. Here, we propose that the observed shape-reversal is related to a surface reconstruction process that affects the binding of surfactants and levels subsequent particle growth, and we provide the first experimental evidence of (1 × 2) missing row surface reconstruction in solution grown metal nanoparticles. Surface reconstruction denotes the relaxation process of the first few atomic layers of a metal surface in the vertical direction to optimize the electron density profile of the surface.45 One of the best studied reconstruction processes is the (1 × 2) missing row reconstruction in face-centered cubic metals whereby a (110) surface reconstructs by elimination of alternating rows of atoms in the [110] direction into a (111) (43) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; Springer Series in Materials Science; Springer: New York, 1995; Vol. 25. (44) Link, S.; Mohamed, M. B.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 3073. (45) Phillies, R. Crystals, Defects and Microstructures; Cambridge University Press: Cambridge, 2001.

a During reconstruction, every second nearest neighbor row is removed, resulting in the formation of (111) microfacets. (a) Frontview of the reconstruction process and (b) side-view of a reconstructed surface. Numbers denote atomic layers.

microfaceted surface. The reconstruction process is illustrated in Scheme 3. Missing row reconstruction has been extensively studied in planar metal surfaces and has recently been reported to occur in electrochemically grown gold nanowires with {110} side facets.46-48 Figure 11 depicts high-resolution electron micrographs of a nanorod oriented along the [110] direction, revealing the characteristic (111) “microfacets” that are typical for the (1 × 2) missing row reconstruction process. For planar surfaces, the reconstruction process has been described by the Avrami equation, that is, N(t)/N∞ ) 1 - exp(-ktn), where N(t) denotes the number of atoms participating in the reconstruction at time t, N∞ is the number of surface sites, n ≈ 1 is the Avrami coefficient, and k is a parameter depending on the temperature as well as the activation energy for atomic motion.47,49,50 Assuming that reconstruction processes of nanoparticle surfaces in solution obey similar relations, it is natural to conclude that a nanorod (initially formed with {110} facets) will undergo a surface reconstruction that will alter the packing density and energy of the particle’s facets and result in conditions that will favor isotropic growth. Surface reconstruction might thus constitute a competing kinetic process that naturally gives rise to an “optimum time scale” for rod formation. Future experiments could probe the implications of surface reconstruction on particle growth by comparing the growth kinetics of single-crystalline and 5-fold twinned nanorod morphologies. Since in the latter case side facets are constituted by the lower energy gold {100} facets, the kinetics of surface reconstruction is likely to be reduced.

Conclusion Gold nanorods prepared by the established seeded-growth method can be used as seed-crystals for the preparation of a variety of modified particle shapes such as low-anisotropy rods or Ξ-type particles by selective overgrowth using modified growth conditions that promote the growth of particular crystal facets. The presented results support earlier reports of the mostly singlecrystalline structure of seed nanocrystals and nanorods prepared using the Ag(I)-mediated seeded-growth procedure. For all conditions tested in our study, the average rod anisotropy was found to decrease during overgrowth, indicating that initial rod (46) Wrigley, J. D.; Ehrlich, G. Phys. ReV. Lett. 1980, 44, 661. (47) Penka, V.; Behm, R. J.; Ertl, G. J. Vac. Sci. Technol. 1986, 4, 1411. (48) Wang, Z. L.; Gao, R. P.; Nikobakht, B.; El-Sayed, M. A. J. Phys. Chem. B 2000, 104, 5417. (49) Avrami, M. J. Chem. Phys. 1939, 7, 1103. Avrami, M. J. Chem. Phys. 1940, 8, 212. Avrami, M. J. Chem. Phys. 1939, 9, 177. (50) The Avrami parameter n ≈ 1 was determined in ref 46 for a system comprising planar Ni(110) in the presence of gaseous hydrogen as adsorbent. Thus, n might be somewhat different in the present case.

Structural EVolution of Gold Nanorods

Langmuir, Vol. 23, No. 20, 2007 10315

Figure 11. High-resolution electron micrograph of a (110) oriented nanorod revealing the (1 × 2) missing row surface reconstruction process of {110} facets. Panel a: nanorod. Panel b: profile view of surface; arrows indicate characteristic “{111} microfacets” due to the absence of atomic rows along the {110} facets. Panel c: schematic view of atom positions in micrograph of panel (b).

formation depends on a subtle balance of kinetic parameters. The observation of the initial increase and subsequent decrease of particle anisotropy can be rationalized as the consequence of competing surface reconstruction processes that result in lowerenergy surface morphologies and in alterations of the binding of ligands to the particle surface. This finding suggests that strategies for the synthesis of high aspect ratio nanorods could capitalize on the presence of ligands or additives that impede the reconstruction process. It also provides a rationale for the relevance of Ag(I) that was shown to be critical to retain particle

anisotropy during the overgrowth reaction, since reconstruction processes on plane metal surfaces were shown to be dramatically affected by the presence of secondary metal atoms. Acknowledgment. Financial support by the German Science Foundation (Bo 1948-1/2) and the National Science Foundation (CTS-0521079) is gratefully acknowledged. The authors thank Prof. Dr. J. Mayer and F. Dorn at the GFE at RWTH Aachen University for their help with high-resolution electron microscopy. LA7015325